一种基于运动区域的姿势识别方法  

Motion-region-based Gesture Recognition Method

作　　者：王兴伟[1] 才国清[1] 刘积仁[2] 大西晟[3] 

机构地区：[1]东北大学计算中心,辽宁沈阳110004 [2]东北大学软件中心,辽宁沈阳110004 [3]名古屋大学信息工程系,日本国名古屋4648601

出　　处：《东北大学学报（自然科学版）》2001年第1期5-8,共4页Journal of Northeastern University(Natural Science)

基　　金：国家'九五'科技攻关项目! (96 B0 8 2 )

摘　　要：提出一种基于运动区域的姿势识别方法 ,详细讨论了一种两级姿势建模架构及其与之相对应的两级姿势识别架构 ,即为姿势建立了两级模型 :SM (SimpleModel)模型 (简单模型 )和FD(FourierDescriptors)模型 (Fourier描述子模型 ) ,先使用SM模型后使用FD模型进行识别·描述了基于该方法而实现的一种姿势识别系统的体系结构及其主要构件的功能 ,简要介绍了该系统基于SGIO2工作站与IRIX 6 3操作系统的原型实现·实验结果表明 ,对几种典型姿势的识别率一般可以达到A motion region based gesture recognition approach was presented. A two level modeling method for gestures and the corresponding recognition method were investigated. That is, a two level model SM(simple model) and FD(fourier descriptors) was established; the SM model was used to recognize gestures,and then FD model was used for identification. The gesture recognition system architecture was described based on the proposed approach, and its main components were introduced. In addition, the prototype system was described briefly. This system is based on SGI O2 workstation and IRIX 6 3 operating systems. The recognition rate of typical kinds of gestures is above 87% experimentally.

关 键 词：姿势识别 运动区域 建模 SM模型 FD模型 

分 类 号：O235[理学—运筹学与控制论]